1 |
Michael Orlitzky <mjo <at> gentoo.org> writes: |
2 |
|
3 |
|
4 |
|
5 |
> > I should qualify that -- a lot of the descriptions suck, not all of |
6 |
> > them. When in doubt, I let the profile decide. |
7 |
|
8 |
Thanks for all the deprecated flag cleaning tools/ideas. |
9 |
|
10 |
|
11 |
> Nah, most of them suck. USE=derp enables libderp? Awesome. WTF does |
12 |
> libderp actually DO for this package? We need a policy change to force |
13 |
> developers to document all of their USE flags, but nobody wants to do |
14 |
> the work I guess. (It would be hard.) |
15 |
|
16 |
Yep. Digging a bit deeper into the ebuilds and codes does time limit |
17 |
one's ability to discern what a particular flag is actually doing |
18 |
in a given ebuild, particularly when used in multiple builds. |
19 |
|
20 |
> First, you'd have to update euse (or a similar tool) to make the |
21 |
> information available. There's no point in adding nice descriptions to |
22 |
> metadata.xml if the users can't query it. |
23 |
|
24 |
> Then, you'd have to figure out all the exceptions. Eclasses can add |
25 |
> things to IUSE, so you would have to figure out the right language, like |
26 |
> "you must document every flag in IUSE in the *ebuild*". You'd also want |
27 |
> exceptions in cases like security bugs -- if there's a vulnerability in |
28 |
> PHP tomorrow, you don't want to force me to add descriptions for 50 USE |
29 |
> flags before I can fix it. Maybe just grandfather in existing versions. |
30 |
|
31 |
Thanks for the insight and example. This is the sort of insight that belongs |
32 |
in the devManual, imho. The devManual often does not explain the 'big' |
33 |
picture and thus becomes a collections of facts, until one is bitten, goes |
34 |
back and reads portions of the devManaul and then has an 'ah-ha' moment. |
35 |
|
36 |
> Repoman needs to enforce it, or else people will ignore it. |
37 |
|
38 |
Excellent idea. I know that Gokturk has been working on the dev manual, and |
39 |
doing a fantastic job, imho. He is quite friendly and receptive to ideas. I |
40 |
just think he has keen insight in organizing and documents, so that is why I |
41 |
mention him here. He seems to be sufficiently 'diplomatic' to mix all the |
42 |
ideas and come up with a brew that is pleasing to most. Maybe this repoman |
43 |
enhancement is a good topic for discussion in gentoo-dev, as repoman is |
44 |
under study by quite a few folks these days? Surely the devManual and |
45 |
repoman should be fully consistent and compatible resources.... |
46 |
|
47 |
> Finally, you'd have to push it through the council and fight about it on |
48 |
> the mailing lists for a while. |
49 |
|
50 |
|
51 |
We'll be pulling for you (MO) on additional vision and upgrades to |
52 |
repoman.... gentoo-dev is quite wonderful when folks 'go at it'. It keeps us |
53 |
calm on gentoo-user, since the devs usually have sharps as insight once |
54 |
ideas start flying around.... |
55 |
|
56 |
|
57 |
Thanks to all for the insights, |
58 |
James |